Effect of Large System Latency of Virtual Auditory Display on Listener's Head Movement in Sound Localization Task

Virtual Auditory Display (VAD) technology is expected to enable the development of new communication tools and many other related applications. However, in computer-network-based communications, large latencies can sometimes occur. Therefore, the influence of large system latency (SL), up to 2 s, on VAD-based sound localization tasks was investigated in terms of the precision and time course of sound localization performance by listeners engaged in head movements. A software VAD system developed by the authors on a Linux PC (with SL of 12 ms) was used in the experiments. Listeners were asked to indicate the location of a virtual sound source by moving their heads in order to face the direction of the perceived sound image. Virtual sound sources were presented to the listeners with one of seven amounts of system latency (12, 50, 100, 200, 500, 1000 and 2000 ms). While the latency detection threshold has been estimated as an SL of about 75 ms, no significant influence on accuracy of sound localization was observed for any of the tested SLs. On the other hand, the time to conclude the sound localization increased as the SL increased. Moreover, a remarkable overshoot was observed in the listener’s head movement particularly when SL was greater than 500 ms. This strongly suggests that the tolerable SL caused by network communications should be kept smaller than 500 ms for VAD applications. [